Privacy-preserving architecture for providing feedback to clinicians on their clinical performance

Abstract

Background: Learning from routine healthcare data is important for the improvement of the quality of care. Providing feedback on clinicians' performance in comparison to their peers has been shown to be more efficient for quality improvements. However, the current methods for providing feedback do not fully address the privacy concerns of stakeholders. Methods: The paper proposes a distributed architecture for providing feedback to clinicians on their clinical performances while protecting their privacy. The indicators for the clinical performance of a clinician are computed within a healthcare institution based on pseudonymized data extracted from the electronic health record (EHR) system. Group-level indicators of clinicians across healthcare institutions are computed using privacy-preserving distributed data-mining techniques. A clinician receives feedback reports that compare his or her personal indicators with the aggregated indicators of the individual's peers. Indicators aggregated across different geographical levels are the basis for monitoring changes in the quality of care. The architecture feasibility was practically evaluated in three general practitioner (GP) offices in Norway that consist of about 20,245 patients. The architecture was applied for providing feedback reports to 21 GPs on their antibiotic prescriptions for selected respiratory tract infections (RTIs). Each GP received one feedback report that covered antibiotic prescriptions between 2015 and 2018, stratified yearly. We assessed the privacy protection and computation time of the architecture. Results: Our evaluation indicates that the proposed architecture is feasible for practical use and protects the privacy of the patients, clinicians, and healthcare institutions. The architecture also maintains the physical access control of healthcare institutions over the patient data. We sent a single feedback report to each of the 21 GPs. A total of 14,396 cases were diagnosed with the selected RTIs during the study period across the institutions. Of these cases, 2924 (20.3%) were treated with antibiotics, where 40.8% (1194) of the antibiotic prescriptions were narrowspectrum antibiotics.

The Norwegian PraksisNett: a nationwide practice-based research network with a novel IT infrastructure

“…Importantly, in the Snow system, no central database is required, as each general practice stores data within its own IT infrastructure, warranting local control of processing and ensuring the anonymity of sensitive patient information [ 38–40 ]. Privacy by design principle, extensive risk assessment and risk mitigation measures have been applied to reduce the risks [ 39 , 40 ]. As the Snow system only handles pseudonymized and encrypted patient data the risks from using the system, related to being a data controller (General Data Protection Regulation (GDPR) responsibility) is minimized.…”

“…The Snow system can perform the basic statistical analyses for power calculations. The local internal storage also supports practice-internal quality improvement work [ 39 ]. Data extraction tools and software are a part of the Snow system, which has been operational since 2010 [ 40 ].…”

A Privacy-Preserving Audit and Feedback System for the Antibiotic Prescribing of General Practitioners: Survey Study

“…Access to the feedback report is restricted to the respective GP and is provided through a web client or email as an encrypted PDF file, and the decryption key is sent to a mobile phone. We refer interested readers to a study by Yigzaw et al [ 13 ] for an elaborated description of the system.…”

“…This study evaluates a scalable, privacy-preserving A&F system [ 13 ] in clinical settings. The system was deployed in three GP offices and used to generate feedback for 20 GPs on their antibiotic prescribing for selected respiratory tract infections (RTIs), which was viewed in comparison to the average performance of peers.…”

Privacy-preserving association rule mining based on electronic medical system

“…Privacy-preserving association rule mining for horizontally partitioned healthcare data [ 19 ] and vertically partitioned healthcare data [ 20 ] were both proposed by Domadiya and Rao respectively. Yigzaw et al [ 21 ] proposed a feasible architecture which can protect the privacy of the patients, clinicians, and healthcare institutions as well as mine the clinical performance of a clinician over the patient data from EHR systems.…”